Closed froesef closed 3 years ago
Thanks for the PR, could you explain a bit why is this needed? For confluent Registry their client should be enough, or not? Or is there some other registry that you need to access?
We have an internal registry that is not fully confluent compatible. Also we're using different protocols than the standard CachedSchemaRegistryClient
.
Thanks for considering this PR! I did some general changes as well in this PR:
Is it fine if I leave these changes also in this PR or do you wish a separate one for these?
Is it fine if I leave these changes also in this PR or do you wish a separate one for these?
Thanks, these are small changes let's keep them here.
What is reason for settingspark.driver.bindAddress
? Does it improve performance?
If you look at the earlier commits it failed to bind the spark driver to a port when starting the spark context. I'm not exactly sure why this happened now as nothing in my PR remotely touches areas like this so it might be a change in GH actions. Anyways the error message said to set a binding address so I did. This is not related to performance. (the UI part is saving like 10 seconds in the Spark context startup time)
Thanks, @froesef for your contribution!
This PR proposes to support a custom registry client that can be specified in the config map that is configured with
usingSchemaRegistry(config: Map[String, String])
.This enables custom registry clients that implement different protocols or authentications. Configuration of these clients can be done by the same
Map[String, String]
.An example custom registry (with no actual implementations of the functions) can be found in
MyRegistry
.